
TO DO's for proj, rel 4:

It is difficult, if not impossible, for a program to be all things for
all users.  Proj was designed to be a batch process filter for carto-
graphic transformations with emphasis upon providing precise results
at a reasonable speed.  Many details, such as those related to input
and output coordinate system, which tend to be highly specific and
limited in scope are purposefully ignored.  It is felt that they are
better handled by external pre- and post-processes or details of
runtime usage.  Exceptions will have to be made for options which
relate to the internal mathematics of the projection and cannot be
handle in any other way.

          ----------------------------------------

The following is a list of proposed changes to be made to proj.
They are not in priority order:

ANSIize geod and include in package. DONE

SPCS scripts into package. DONE

Mean radius latitude from specified ellipsoid: +Rm= (DONE)

Common setup for conics (Conformal modified, other really OK).

Add k_0 factor for all(?) projections.  Also change all current "k"s
	to k_0 (with old alternative, for now). (PARTIAL)
	Update note: k or k_0 only applicable to conformal projections.
	(k_0 added to Lambert Conformal and Mercator anything else can
	wait)

Add Russian variant of UTM: GK

Fix both UTM and UPS to REQUIRE ellipsoid.  Inconsistent now (DONE).

Benchmark forward<->inverse tables for at least all ellipsiodal projections
	used in plane coordinates systems accurate to 1mm/.00003".  These
	should be acceptable "standards" by which to judge a projection's
	accuracy performance.  There are two reasons for wanting at least
	this precision: generating approximation functions and GPS can,
	with time or DOD secrets, give these results.
	NOTE: such tests have been made with SPCS procedures and compared
	with USGS/GCTP-II.

	Problem: finding some one or a group to stick their neck out on certifying
	the numbers.

	One of the biggest problems I have is getting test values of reliable
	precision.  Even Snyder goofed up on his example for Oblq. Space Merc.
	and it took me days, plus correspondence to resolve the problem.

New projections.  This is a tentative list and a better idea of what is
	needed may be suggested by DMA manual (hopefully will receive a copy
	mid Nov.).

	a. Add all of Snyder's modified stereographics. Easy. (DONE)
	b. Redo and fixup Chamberlin Trimetric as well as do inverse (messy).
	c. Modified polyconic (Haines)
	d. Modified polyconic (Lallemand, IMC)
	e. Adams' stereographics---need to look into these more.
	   Is anyone using it?
	f. More general form of azimuthal perspective.
	g. Elliptical form of azimuthal equidistant (for Guam).
	h. General elliptical space oblique mercator (groan).
	e. New Zealand Grid Map. (DONE)

Script file similar to NAD conversions that apply to international
grid systems---sounds feasible based upon sampler copies.  DONE

Add convergence and scale based on formulary for projections used in
plane coordinate applications (messy, especially internally).

          ----------------------------------------

The following suggestions are held in contemplation pending further
justification for need and commonality of occurence:

	Grad angular units.  If implemented, it probably should be in a from
		similar to radian suffix now used.
	
	Non-Greenwich geographic base.
	
	General x-y scaling for equirectangular cylindrical seems peculiar choice
		in lieu of normal use of lat_ts.
	
	South Oriented Transverse Mercator.

          ----------------------------------------

Documentation: (Ah, good ol' documentation)

	Pending a complete rewrite of the current manual for Rel.3 I am leaning
	toward a supplement that documents changes and new projections.  One
	problem I am having with the R3 manual is the repeatative nature of
	some sections and trying to design a useful alternative.  Secondly, I
	would like to get it into two column mode to improve readability and
	save paper.  (DONE, two-column doesn't really work)

	Several sections of new material need to be added: accuracy/precision
	demonstations and problems, benchmarking projections, bivariate
	Chebychev approximation technique and a short addendum on the geodesic
	program.

	Some form of testing a porting of the software needs to be supplied
	with the release "tape." (See tests in nad).
